Session Item

To assess long-term stability of machine QA results for the hybrid 1.5T MR-Linac (MRL) during the first year of clinical treatments. For a hybrid system as the Unity MR-Linac (Elekta AB, Stockholm, Sweden), checks regarding accelerator and imaging characteristics have to be monitored in dedicated time intervals.

Routine machine QA at the 1.5T MR-Linac (MRL) was performed using the following protocol. Dose calibration in water (Stationary Water Phantom MR, PTW-Freiburg, Germany) using a Farmer type chamber (PTW 30013), beam quality factor Q (ratio of dose measured in 20 cm vs. 10 cm water depth) and a Periodic Image Quality Test (PIQT) with a special MR phantom (Philips Oy, Vantaa, Finland) for basic MR system performance, including Signal-to-Noise Ratio (SNR) and slice profile measurements evaluating image intensity with respect to Full Width at Tenth Maximum (FWTM), are checked weekly. The accordance of MR and accelerator isocenters (MR-to-MV) using a dedicated phantom (Elekta), a check of electronic portal imaging device (EPID) central pixel position to map the MV isocenter stability (MV Alignment Phantom, Elekta), dose profile measurements for field sizes of 2x2 cm², 4x4 cm², 10x10 cm², 30x20 cm² and 40x22 cm² (STARCHECK maxi, PTW-Freiburg) and a test for geometric accuracy of MR images (3D Geometry QA Phantom, Philips), are verified monthly. The radius of the MV isocenter circle is calculated with a Winston-Lutz-test quarterly.

In this work, the QA results for weekly, monthly and quarterly checks over the first year of operation are analyzed and presented.

Fig. 1 shows an overwiev of the weekly dose calibration with majority of values within a range of ± 1%. Absolute dose checks resulted in a mean dose difference with respect to the dose defined during calibration of (0.21 ± 0.32)%, Q was 0.705 ± 0.001. For PIQT the SNR reached values of 71.38 ± 1.25 (TE=30ms) and 53.99 ± 0.92 (TE=100ms), specification SNR > 59 or 44, respectively. The slice profile measurements led to values of FWTM = (7.74 ± 0.51) mm, specification < 8 mm. Initial MR-to-MV checks indicated isocenter shifts of up to Δx = 0.37 mm, Δy = 0.10 mm and Δz = 0.80 mm, indicating loose fixation of the EPID panel. After re-fixation of the panel holder, the shifts were reduced to Δx = (0.05 ± 0.02) mm, Δy = (0.04 ± 0.04) mm and Δz = (0.02 ± 0.01) mm. The deviation of the central EPID pixel was found to be (0.04 ± 0.06) pixel and (0.11 ± 0.09) pixel, respectively. The radius of the isocenter circle was (0.300 ± 0.137) mm (tolerance < 0.500 mm). Dose deviations concerning the reference measurements for all field sizes were below 1%. An overview of the geometric accuracy of the MRI is shown in Fig. 2.

The presented QA data for the Linac and the MR part as well as the tests for the hybrid MR-Linac system showed a high level of stability and robustness. QA tests established on a regular weekly, monthly or quarterly basis seem to be adequate for routine machine QA of a clinical MR-Linac system.